Until quite recently, the consideration of oligonucleotides in any capacity other than strictly informational was unheard of. Despite the fact that certain oligonucleotides were known to have interesting structural possibilities (e.g., t-RNAs) and other oligonucleotides were bound specifically by polypeptides in nature, very little attention had been focused on the non-informational capacities of oligonucleotides. For this reason, among others, little consideration had been given to using oligonucleotides as pharmaceutical compounds.
There are currently at least three areas of exploration that have led to extensive studies regarding the use of oligonucleotides as pharmaceutical compounds. In the most advanced field, antisense oligonucleotides are used to bind to certain coding regions in an organism to prevent the expression of proteins or to block various cell functions. Additionally, the discovery of RNA species with catalytic functions--ribozymes--has led to the study of RNA species that serve to perform intracellular reactions that will achieve desired effects. And lastly, the discovery of the SELEX process (Systematic Evolution of Ligands by Exponential Enrichment) (Tuerk and Gold (1990) Science 249:505) has shown that oligonucleotides can be identified that will bind to almost any biologically interesting target.
SELEX is a method for identifying and producing nucleic acid ligands, termed "nucleic acid antibodies", e.g., nucleic acids that interact with target molecules (Tuerk and Gold (1990) Science 249:505). The method involves selection from a mixture of candidate oligonucleotides and step-wise iterations of binding, partitioning and amplification, using the same general selection theme, to achieve virtually any desired criterion of binding affinity and selectivity.
The use of antisense oligonucleotides as a means for controlling gene expression and the potential for using oligonucleotides as possible pharmaceutical agents has prompted investigations into the introduction of a number of chemical modifications into oligonucleotides to increase their therapeutic activity and stability. Such modifications are designed to increase cell penetration of the oligonucleotides, to stabilize them from nucleases and other enzymes that degrade or interfere with the structure or activity of the oligonucleotide analogs in the body, to enhance their binding to targeted RNA, to provide a mode of disruption (terminating event) once sequence-specifically bound to targeted RNA and to improve their pharmacokinetic properties.
Recent research has shown that RNA secondary and tertiary structures can have important biological functions (Tinoco et al. (1987) Cold Spring Harb. Symp. Quant. Biol. 52:135; Larson et al. (1987) Mol. Cell. Biochem. 74:5; Tuerk et al. (1988) Proc. Natl. Acad. Sci. USA 85:1364; Resnekov et al. (1989) J. Biol. Chem. 264:9953). PCT Patent Application Publication WO 91/14436, entitled "Reagents and Methods for Modulating Gene Expression Through RNA Mimicry," describes oligonucleotides or oligonucleotide analogs which mimic a portion of RNA able to interact with one or more proteins. The oligonucleotides contain modified internucleoside linkages rendering them nuclease-resistant, have enhanced ability to penetrate cells, and are capable of binding target oligonucleotide sequences.
Although there has been a fair amount of activity in the development of modified oligonucleotides for use as pharmaceuticals, little attention has been paid to the preparation and isolation of these compounds on a scale that allows clinical development. The conventional laboratory scale 1 .mu.mole automated oligonucleotide synthesis does not provide a sufficient amount of the compound of interest to enable clinical development. For clinical development oligonucleotides must be produced in gram-scale to multigram scale amounts at a minimum. Although there are reports of large-scale oligoribonucleotide syntheses in the literature, the term "large-scale" has been applied to the 1 to 10 .mu.mole scale, rather than gram-scale or kilogram-scale amounts. (Iwai et al. (1990) Tetrahedron 46:6673-6688).
The current state of the art in oligonucleotide synthesis is automated solid phase synthesis of oligonucleotides by the phosphoramidite method, which is illustrated in Scheme 1. (Beaucage and Iyer (1992) Tetrahedron 48:2223-23 11; Zon and Geiser (1991) Anti-Cancer Drug Design 6:539-568. Matteucci and Caruthers (1981) J. Am. Chem. Soc. 103:3185-3191). Briefly the 3'-terminal nucleoside of the oligonucleotide to be synthesized is attached to a solid support and the oligonucleotide is synthesized by addition of one nucleotide at a time while remaining attached to the support. As depicted in Scheme 1 a nucleoside monomer is rotected (P.sub.1) and the phosphoramidite is prepared (1). The phosphoramidite (referred to as the 5'-protected monomer unit) is then covalently attached to the growing oligonucleotide chain (2), via a phosphite triester linkage, through the 5'-hydroxy group of the ribose ring of the growing oligonucleotide chain to yield the oligonucleotide product (3), in which the majority of the growing oligonucleotide chain has been extended by one nucleotide, but a significant percent of chains are not extended. The product (3) is then oxidized to yield the phosphate triester (4). Prior to the addition of the next base to the growing nucleotide chain, the 5'-hydroxyl group must be deprotected. As can be seen in Scheme 1 (compound 4), however, not all of the reactive sites on the solid support react with the 5'-protected monomer. These unreacted sites (referred to as failure sequences) must, therefore, be protected (referred to as capping) (5) prior to deprotection of the 5'-hydroxyl group (6). Subsequent monomers, which have also been protected and converted to the phosphoramidite, are then sequentially added by coupling the 5'-end of the growing oligomer to the 3'-end of the monomer. Each coupling reaction extends the oligonucleotide by one monomer via a phosphite triester linkage. At each step--and in the case of the initial reaction with the solid support--there are reactive sites that fail to react with the 5'-protected monomer, which results in oligonucleotides that have not been extended by one nucleotide monomer (failure sequences). When the synthesis is complete the desired oligonucleotide (6 (n+1 sequence)) is deprotected and cleaved from the resin, together with all of the failure sequences (n, n-x).
The yield of conventional solid phase oligonucleotide synthesis decreases exponentially with the number of monomers coupled. This increases the difficulty of purifying the crude product away from the failure sequences. Additionally, even after high resolution purification has been achieved, it remains very difficult to verify the sequence and composition of the product, especially if it contains non-standard nucleotides. ##STR1##
Automated oligonucleotide synthesis on solid supports is very efficient for the preparation of small amounts, 0.001 to 0.01 mmol, of a variety of sequences in a minimum amount of time with reasonable yield. It is, however, a highly inefficient process in terms of overall process yield based on input monomer. Typically a 16 fold excess of phosphoramidite is necessary per monomer addition. It has been recognized that the automated solid phase synthesis approach does not readily lend itself to be scaled to a level that allows efficient manufacture of oligonucleotide pharmaceuticals. (Zon and Geiser (1991) Anti-Cancer Drug Design 6:539-568).
The inefficiency of the solid phase synthesis is created to a large extent by the heterophase monomer coupling reaction and by the covalent attachment of both unreacted failure sequences and reaction product to the same support bead. In each cycle, 1-5% of the nucleotide bound to the support does not react with the activated monomer. These unreacted compounds, referred to as failure sequences, as discussed above, must be blocked or capped in order to prevent the subsequent addition of monomers to incomplete oligonucleotides. The generation of failure sequences at every step of the synthesis produces a crude product contaminated with highly homologous byproducts, which must be carried through to the final crude product (see Scheme 1, structure 6 (n, n-x)). As a result, purification of crude synthetic oligonucleotides to a state acceptable for clinical studies is extremely cumbersome and inefficient. To minimize the percent of failure sequences, a large excess of monomer (approximately 16 fold) is used.
A method to scale-up solid phase oligonucleotide synthesis using a higher loaded polystyrene support was reported by Montserrat et al. (1994) Tetrahedron 50:2617-2622. This method, however, does not overcome the primary problem associated with solid phase synthesis, in that a considerable monomer excess is still required to minimize failure sequences. Additionally, the method does not provide consistently satisfactory yields.
In an attempt to decrease the excess of monomer needed to achieve coupling and to achieve easy scaleability, Bonora et al. (1993) Nucleic Acids Res. 21:1213-1217, have investigated using polyethylene glycol (PEG) as a 3'-support that is soluble in the monomer coupling reaction. This method has been used to prepare oligonucleotides by phosphoramidite coupling, H-phosphonate condensation and phosphotriester condensation. (See Bonora (1987) Gazzetta Chimica Italiana 117:379; Bonora et al. (1990) Nucleic Acids Res. 18:3155; Bonora et al. (1991) Nucleosides & Nucleotides 10:269; Colonna et al. (1991) Tetrahedron Lett. 32:3251-3254; Bonora and Scremin (1992) Innovation Perspect. Solid Phase Synth. Collect. Pap., Int. Symp.. 2nd, "Large Scale Synthesis of Oligonucleotides. The HELP Method: Results and Perspectives", pp. 355-358, published by Intercept, Andover, UK; Scremin and Bonora (1993) Tetrahedron Lett. 34:4663; Bonora (1995) Applied Biochemistry and Biotechnology 54:3; Zaramella and Bonora (1995) Nucleosides & Nucleotides 14:809). The weakness of this approach is the unacceptably low recovery of support bound oligonucleotide after each reaction step. Additionally, this method does not address the problem of failure sequences that must be capped and carried through to the final product.
A polyethylene glycol-polystyrene copolymer support has also been used for the scale-up of oligonucleotide synthesis. (Wright et al. (1993) Tetrahedron Lett. 34:3373-3376). At the 1 mmol scale a 96.6% coupling efficiency per monomer addition was reported for an 18 mer DNA. Again, this method does not address the problem of failure sequences bound to the resin.
Zon et al. have suggested a block approach to the synthesis of oligonucleotides, in which a library of dimer or multimer oligonucleotide fragments are prepared in solution and then coupled to each other. (Zon and Geiser (1991) Anti-Cancer Drug Design 6:539-568). The fragments are activated for coupling by differential 5'-deprotection and 3'-phosphorylation. Phosphotriester coupling has been suggested for fragment preparation. (Bonora et al. (1993) Nucleic Acids Res. 21:1213-1217). Due to the comparatively low yield of phosphotriester coupling this approach has not been widely accepted.
In conventional oligonucleotide synthesis, the 5'-protecting group serves to prevent reaction of the 5'-hydroxyl group of one monomer with the phosphoramidite group of a second monomer during the coupling step. The 4,4'-dimethoxytrityl (DMT) group is commonly used as the 5'-protecting group (Schaller et al. (1963) J. Am. Chem. Soc. 85:3821) of the 5'-protected monomer unit added during oligonucleotide synthesis. This group is chosen because of the ease and selectivity with which it can be removed from the 5'-oxygen of the oligonucleotide product prior to addition of the next 5'-protected monomer unit (for a review see: Beaucage and Iyer (1992) Tetrahedron 48:2223-2311). In solution, deprotection of the 5'-DMT group is impaired by the reversibility of acid induced detritylation. In order to drive this reaction to completion, a scavenger of the free trityl cation is added for solution-phase detritylation (Ravikumar et al. (1995) Tetrahedron Lett. 36:6587). It has been recognized that the final 5'-terminal DMT group may serve as a hydrophobic handle which allows separation of the full-length product oligonucleotide from shorter failure sequences by reverse phase chromatography. Additionally, highly hydrophobic analogs of the DMT group have been prepared to enhance the resolution of the separation of full length deprotected oligonucleotide product from failure sequences after complete solid phase synthesis (Seliger and Schmidt (1987) Journal of Chromatography 397:141). In another approach, a fluorescent trityl analog has been used for the 5'-terminal protecting group during oligonucleotide synthesis to allow facile detection of full-length product in crude deprotected oligonucleotide (Fourrey et al. (1987) Tetrahedron Lett. 28:5157). Colored trityl groups were devised to allow monitoring of specific monomer additions during solid phase oligonucleotide synthesis (Fisher and Caruthers (1983) Nucleic Acids Res. 11: 1589). Other modified trityl groups have been prepared for the purpose of changing or enhancing the selectivity with which the trityl group can be removed from the oligonucleotide during solid phase oligonucleotide synthesis (for a review, see: Beaucage and Iyer (1992) Tetrahedron 48:2223-2311).
To date, trityl groups which allow anchoring of the product to a resin or membrane during oligonucleotide synthesis in solution have not been designed. Additionally, trityl groups which can covalently react with a derivatized resin, membrane or soluble polymer have not been reported.
The Diels-Alder reaction is a cycloaddition reaction between a conjugated diene and an unsaturated molecule to form a cyclic compound with the .pi.-electrons being used to form the new a bonds. The Diels-Alder reaction is an example of a [4+2] cycloaddition reaction, as it involves a system of 4-.pi. electrons (the diene) and a system of 2-.pi. electrons (the dienophile). The reaction can be made to occur very rapidly, under mild conditions, and for a wide variety of reactants. The Diels-Alder reaction is broad in scope and is well known to those knowledgeable in the art. A review of the Diels-Alder reaction can be found in "Advanced Organic Chemistry" (March, J., ed.) 761-798 (1977) McGraw Hill, N.Y., which is incorporated herein by reference.
To date, although a number of attempts have been made, there still remains a need for a method to produce oligonucleotides in large quantities, in continuous operations, at low cost and without laborious purification.